[MarcusPottinger]

Hi,

I am trying to run an interFoam simulation of a water/air tank stirred at 400RPM by a paddle. The water is at the bottom of the tank and stationary to start, with an air layer above and an open atmosphere inlet/outlet at the top of the tank.

I am using the KEpsilon solver, with blockMesh, snappyHexMesh, and a cyclic AMI approach.

This is similar to the mixerAMI tutorial, so I have cloned the tutorial fV files, and am using my own calculated values for BCs for K and Epsilon based on 1% turbulence intensity.

I have tried a coarse blockMesh with high refinement from snappy, as the blade geometry is highly complex. I have also tried a fine blockMesh with minimal refinement in snappy.

The coarse mesh ran for 1.8 seconds at Cmax 0.5, then diverged and crashed and burned.
The fine mesh immediately diverges.

My mesh is showing a minor skew issue on 1-4 faces depending on mesh quality, but the max skew is <4.5, and the faces are well separated in the geometry so I ignored it.

I have attached my '0', 'fV', transport, turbulence files for ref.

My initial timestep is 0.0001, with maxCo 0.5 to try and get this to run properly.

Can anyone see an issue with this setup?

Thanks!


Edit to add:
When it fails, I get divergence like this (lines omitted for brevity) - Courant shoots up causing DeltaT to fall to e-lots, then the phase volume fractions become physically impossible.

Courant Number mean: 0.000780024 max: 21.17
Interface Courant Number mean: 3.45556e-05 max: 3.31781
deltaT = 2.89475e-11
Time = 4.94633e-05


MULES: Solving for alpha.water
Phase-1 volume fraction = -4.37142 Min(alpha.water) = -6.38818e+12 Max(alpha.water) = 5.21436e+12